This week's 7-day Smoke test reveals GPT-5.5's execution score plummeting while two Claude models stage a dramatic reversal, though stability remains a concern. The data also highlights volatility and integrity rating fluctuations across multiple models.
At 2:30 AM on May 6, 2026, Anthropic's latest AI model, Claude Opus 4.8, experienced a global service crisis, with developers unable to use Claude Code features due to continuous "Service is busy" errors. The outage lasted nearly an hour, causing significant disruptions and losses for developers worldwide.
The results of the Smoke Lite evaluation on June 5, 2026, show that 9 out of 11 models tied at 77.5 on the main leaderboard, forming a rare tie. Their common feature is that all scored a perfect 100 on the Code Execution dimension, but only 50 on the Material Constraint dimension.
Discussions around AI agents have surged on X (formerly Twitter), with participants including developers, investors, and corporate executives. While excitement grows over multimodal integration and enterprise automation, concerns about the gap between hype and practical deployment remain significant.
Alphabet recently announced a massive $85 billion financing round to expand its Google AI business, setting a company record and marking one of the largest single AI investments in the global tech industry.
In 2025, Anthropic filed an IPO prospectus with the SEC, marking the AI industry's transition from early hype to mature commercialization. The move is seen as a bellwether for the sector.
A British MP has filed a lawsuit against xAI, alleging that its chatbot Grok generated sexualized images, igniting intense debates about AI ethics, content safety, and regulatory responsibility.
TSMC's CEO publicly stated that AI chip demand remains robust, driving the company's performance growth and boosting the semiconductor sector. However, challenges such as geopolitical factors and intensified competition persist.
A controversial discussion has emerged in the tech community suggesting that the revenue cycle between Microsoft and OpenAI mirrors the business logic of the dot-com bubble era. This has led to notable stock market fluctuations and renewed scrutiny of AI investment sustainability.
Cloudflare's latest report reveals that traffic generated by AI agents has exceeded human user traffic for the first time, marking a significant shift in the internet landscape and sparking widespread discussion in the tech community.
Smoke's quick test results today clearly show that the code execution dimension is nearly saturated. Ten out of eleven models scored 100, while GPT-5.5 dropped to 50, dragging its main leaderboard score down to 59.99.
Microsoft's Project Solara is a chip-to-cloud device platform designed exclusively for AI agents, not traditional apps. By basing it on AOSP instead of Windows, Microsoft makes a pragmatic bet on a future where interaction shifts from apps to intents.
WDCD Run #146 (2026-06-03) tested 11 frontier models on multi-turn commitment integrity, recording an average instruction decay of 24.7% from Round 1 to Round 3. Claude Opus 4.7, GPT-5.5, and GPT-o3 tied for first place at 70 points with only -10% decay each.
Run #141 data shows that Grok 4 improved by 10.8 points in a single round, GPT-5.5 improved by 9.2 points, while Qwen3 Max plummeted by 10.8 points. The divergence in adherence capabilities has become clearly visible.
The most brutal finding of the WDCD compliance test is that resource constraints crippled all models, with an average score of only 1.7 across 11 models, far below the other four scenarios.
The core findings of the WDCD three-round test are clear: nearly all models scored high in the constraint establishment phase, but after two rounds of interference, over 60% of models completely abandoned their original commitments under direct pressure.
The WDCD commitment test reveals model weaknesses through a three-round dialogue design. Only three out of eleven models scored 70, with Ernie Bot 4.5 forming a clear gap at 50.
Today's Smoke quick test results show Claude Opus 4.7, Claude Sonnet 4.6, and GPT-5.5 all tied for first with a main ranking score of 87.76. The core reason is that all three achieved a perfect 100 on the code execution dimension, while scoring 72.8 on material constraint, triggering a warn signal.
Anthropic officially submitted its IPO application on June 1, 2026. The decision highlights valuation uncertainties, profitability comparisons, and a stark contrast with OpenAI, which remains unlisted, fueling debate over market expectations for AI startups.
Senator Elizabeth Warren has publicly urged the government to close potential loopholes in AI chip exports to China, criticizing possible policy loosening during the Trump era. She calls for stricter controls to prevent Chinese companies from bypassing restrictions through overseas subsidiaries.
